What Is Multiple Realizability?

Direct Answer

Multiple realizability is one of those philosophical ideas that, once you grasp it, seems almost obvious — but its implications are anything but. The thesis is simple: the same mental state can be physically realized in many different ways. Pain in a human, pain in an octopus, and pain in a hypothetical silicon-based extraterrestrial could all be genuine instances of pain even though the underlying physical machinery is completely different in each case. The mental state is multiply realizable.

This sounds harmless until you realize what it rules out. If multiple realizability is true, then pain can't be identical to any particular physical state. It can't be identical to C-fiber firing (the classic example from type-identity theory), because octopuses don't have C-fibers, and if they can feel pain, then pain is something broader than C-fiber firing. The same goes for any candidate physical state you might propose. If the same mental state can be realized by different physical states, then mental states aren't identical to physical states in the straightforward way that type-identity theory proposed.

Hilary Putnam introduced this argument in his 1967 paper "Psychological Predicates" (later republished as "The Nature of Mental States"). It was aimed squarely at the type-identity theory that J.J.C. Smart and U.T. Place had been developing — the view that mental states are identical to brain states in the same way that water is identical to H2O or lightning is identical to electrical discharge. Putnam's argument showed that this analogy breaks down. Water really is H2O everywhere — there's no alternative chemistry that produces the same macroscopic properties. But pain, Putnam argued, could be produced by many different physical substrates. The relationship isn't identity but realization, and realization is a one-to-many relation.

Putnam's Argument Against Type-Identity Theory

The type-identity theory holds that mental state types are identical to physical state types. Not just token-for-token (this particular pain is this particular brain event) but type-for-type (pain as such is C-fiber firing as such). This is a strong claim. It says that the general category "pain" and the general category "C-fiber firing" pick out the same thing.

Putnam's argument proceeds by considering what would have to be true for type-identity to hold. If pain = C-fiber firing, then any creature in pain must have C-fibers firing, and any creature with C-fibers firing must be in pain. But both directions seem false. Octopuses, which have very different nervous systems from mammals, appear to feel pain — they avoid noxious stimuli, show signs of distress, and can learn to associate certain contexts with harmful experiences. If an octopus is in pain but doesn't have C-fibers, then pain isn't identical to C-fiber firing.

The point generalizes. Consider any physical state type you might identify with a mental state. There will always be possible creatures — or even actual creatures — that share the mental state but not the physical state. Mammals, birds, reptiles, fish, cephalopods: all show behavioral and physiological signs of pain, but their nervous systems differ dramatically. If all of these creatures can be in pain, then pain is multiply realized, and type-identity fails.

Putnam also made the argument more vivid with a science-fiction thought experiment. Imagine we encounter an alien species that is intelligent, communicative, and emotionally complex but whose internal physiology is nothing like ours — perhaps they're silicon-based rather than carbon-based. If one of them says "I'm in pain" and behaves the way a pain sufferer behaves, we'd be inclined to say they're really in pain. But their pain isn't realized by C-fiber firing (they don't have C-fibers) or by any neural process we recognize. The mental state is there, but the physical substrate is entirely different.

The conclusion: mental states can't be identified with specific physical state types. They're defined not by what they're made of but by what they do — their causal-functional roles in the organism's psychology.

Functionalism and the Causal Role

This is where functionalism enters. If mental states can't be identified with their physical realizers, what can they be identified with? The functionalist answer: their causal-functional roles. Pain is the state that's typically caused by tissue damage, that causes distress and avoidance behavior, that interferes with concentration, that causes the belief that something is wrong. Any state that plays this role — regardless of its physical composition — is pain.

Functionalism and multiple realizability are natural partners. Multiple realizability says the same mental state can have different physical realizers. Functionalism explains how: mental states are individuated by role, not by realizer, and the same role can be filled by different physical mechanisms. This is analogous to other cases in science. A heart is defined by its function (pumping blood), not by its material composition. An artificial heart, a pig's heart, and a human heart can all be hearts because they all perform the same function, despite being made of different stuff. Similarly, pain is defined by its function in the organism's cognitive economy, and any system that implements that function is in pain.

This picture has enormous appeal. It accommodates the intuition that animals with different nervous systems can share mental states. It makes sense of the idea that a sufficiently advanced computer could genuinely think. It allows psychology to proceed autonomously from neuroscience — psychology studies the functional level, neuroscience studies the implementation level, and the two are related but not identical. This autonomy thesis became central to the architecture of cognitive science: the mind is studied at the computational level, and the neural implementation is a separate (if related) question.

The Connection to Cognitive Science

Multiple realizability has shaped cognitive science profoundly. If mental states are multiply realizable, then the right level of explanation for cognition isn't the neural level but the functional or computational level. You can't understand thinking by looking at neurons alone, any more than you can understand a software program by examining the transistors it runs on. You need to understand the program — the functional organization — and the hardware is just one possible implementation.

This justified the cognitive revolution of the 1950s and 60s, which replaced behaviorism with information-processing models of the mind. Cognitive psychologists study memory, attention, language, and reasoning at the functional level, largely independent of their neural implementation. Neuroscience provides constraints and details, but the primary explanatory framework is computational and functional. Multiple realizability is the philosophical foundation for this division of labor.

It also has implications for philosophy of science. If mental states are multiply realizable, then psychology is a special science with its own laws and explanations that can't be reduced to physics. The relationship between psychology and neuroscience is like the relationship between biology and chemistry or between economics and psychology: each science operates at its own level of description, with its own vocabulary and laws, and reduction to a lower level isn't always possible or useful. This view, called non-reductive physicalism, became the orthodox position in philosophy of mind for several decades.

Objections to Multiple Realizability

Multiple realizability hasn't gone unchallenged. Jaegwon Kim, one of the most influential critics, raised what's known as the causal exclusion problem. If mental states are multiply realizable and thus not identical to any specific physical state, then each token of a mental state is realized by some specific physical state. But that physical state already has a sufficient physical cause. So what causal work is left for the mental state to do? Either the mental state is causally redundant (because the physical state does all the causal work) or it causes things in its own right (which seems to violate the causal closure of the physical). Kim argued that this tension pushes us back toward reduction: if mental states are to be causally efficacious, they may need to be identified with their physical realizers after all.

More direct empirical challenges have come from William Bechtel and Larry Shapiro, who argue that the actual evidence for multiple realizability is weaker than philosophers have assumed. They point out that convergent evolution often produces similar neural mechanisms in different species — the camera-like eye has evolved independently multiple times, and the underlying molecular machinery is remarkably conserved. If nervous systems converge on similar solutions to similar problems, then the space of possible realizers for a given mental state may be narrower than the thought experiments suggest. Perhaps pain in mammals, birds, and cephalopods is realized by more similar physical mechanisms than we'd expect — not identical, but not as different as the multiple-realizability argument needs them to be.

Shapiro has argued that once we look at the actual constraints on implementing a given function, the number of viable realizers shrinks dramatically. A system that detects damage and triggers avoidance needs certain structural features — sensors, a processing mechanism, a motor output pathway — and these features constrain the physical implementation more tightly than the abstract thought experiments suggest. The octopus and the human may not share C-fibers, but their pain systems may be more physically similar than Putnam's argument assumes, which weakens the case for treating the mental and physical as entirely independent levels.

Carl Gillett has argued that the debate has been muddied by confusion about what "realization" means. He distinguishes between flat realization (where a single property realizes another) and dimensioned realization (where multiple properties across levels jointly realize a property). On the dimensioned view, the relationship between mental and physical is more complex than a simple one-to-many mapping, and this complexity may dissolve some of the philosophical puzzles while preserving the core insight that mental states aren't identical to single physical state types.

Significance and Legacy

Despite the challenges, multiple realizability remains one of the most important ideas in philosophy of mind. Even if the thesis needs qualification — even if the space of actual realizers is narrower than originally imagined — the core insight survives: you can't read off the mental from the physical in any simple way. Two systems can be physically different yet psychologically similar, and this fact has consequences for how we do science, how we understand other minds, and how we think about the relationship between ourselves and the machines we build.

The argument also illustrates something important about philosophical method. Putnam's case relied heavily on thought experiments — imagining octopuses, aliens, and silicon brains. The critics respond with empirical evidence — looking at what actual nervous systems are like. This tension between armchair philosophy and empirical science is itself revealing: the philosophy of mind can't be done purely from the armchair, but it also can't be done purely from the lab. The deepest questions about minds require both conceptual clarity and empirical sensitivity.

Putnam himself, characteristically, didn't stay settled on the view. He moved away from functionalism in the 1980s, arguing that it faced problems about meaning and reference that it couldn't solve. But the argument from multiple realizability outlived his commitment to the theory it was designed to support. It has become part of the standard toolkit of philosophy of mind — a constraint that any theory of mental states must accommodate, whether it's functionalism, identity theory, or something else entirely.

Further Learning

1. Stanford Encyclopedia of Philosophy, Multiple Realizability.
2. Hilary Putnam, "Psychological Predicates," in Art, Mind, and Religion, eds. W.H. Capitan and D.D. Merrill (University of Pittsburgh Press, 1967).
3. Jaegwon Kim, "Multiple Realization and the Metaphysics of Reduction," Philosophy and Phenomenological Research 52 (1992): 1-26.
4. Larry Shapiro, The Mind Incarnate (MIT Press, 2004).
5. William Bechtel and Jennifer Mundale, "Multiple Realizability Revisited: Linking Cognitive and Neural States," Philosophy of Science 66 (1999): 175-207.

To explore related concepts, consider reading about the philosophy of mind, philosophy of science, and the work of Hilary Putnam.